You are on page 1of 5

Performance of MIL-STD 188-184 in Support of IP Traffic

Over UHF Satcomm Dedicated Channels


Yosry A. Barsoum
Gus Amouris

Rich Condello

The MITRE Corporation


145 Wyckoff Road
Eatontown, NJ 07724

PEO C3S
PM MILSATCOM
Fort Monmouth, NJ 07703

Abstract- Ultra High Frequency (UHF) tactical satellite


communications (i.e. UHF TACSAT) will be utilized to
provide range extension for the terrestrial components
of the lower echelon tactical Internet (TI) for the first
digitized division (FDD). The terrestrial components
include Single Channel Ground radio system
(SINCGARS) and Enhanced position location system
(EPLRS). Because of the limited UHF TACSAT
resources (i.e. number of channels) and the nature of
the lower echelon TI traffic (i.e. relatively short packet
sizes and multi-destined), a multi-access protocol is
required to allow for multiple users to make efficient
use of the UHF TACSAT channel. The UHF TACSAT
multi-access capability is governed by MIL-STD 188184.
In this paper, a qualitative as well as a
quantitative assessment of the MIL-STD 188-184
performance for passing IP traffics over UHF
TACSAT dedicated channels is presented.

Access (DAMA) operations. The mechanism that allows


Internet Protocol (IP) multi-access on UHF dedicated
TACSAT channels is the MIL-STD 188-184.

UHF TACSAT

Lower Echelon
TI EPLRS
Backbone
Lower Echelon TI EPLRS
Backbone

PLT SIP
SINCGARS
Net

BN SIP
SINCGARS
Net

Lower Echelon
TI EPLRS
Backbone
PLT SIP
SINCGARS
Net

CO SIP
SINCGARS
Net

PLT SIP
SINCGARS
Net

Lower Echelon
TI EPLRS
Backbone
PLT SIP
SINCGARS
Net

1. Introduction
Evolving warfighting doctrine and battlefield
digitization are having a direct impact on the
communications architecture. Battlefield digitization is
being relied upon to improve the Warfighting capability of
U.S. Forces, allow for a reduced force structure, and
enable a given force to cover large areas. This places an
increased reliance on beyond-line-of-sight (BLOS)
communications. Additionally, the force structure is
becoming more mobile which places a greater demand for
BLOS on the move communications. UHF TACSAT
provides both BLOS communications and offers the
flexibility for communications on the move.
The current lower echelon TI consists of SINCGARS
stub networks and EPLRS as the network backbone. The
UHF TACSAT will be used to extend the range of the TI
as illustrated by Figure 1. The traffic characteristics in the
lower echelon TI largely consist of situation awareness
(SA) and command and control (C2) messages. SA traffic
is primarily broadcast in nature, while the large majority of
the C2 traffic is multicast. Because of the multidestination nature of the traffic (i.e. SA and C2 multicast),
combined with the limited UHF resources, UHF TACSAT
channels will have to be shared among multiple users for
both dedicated as well as Demand assigned Multiple

Figure 1. Notional Architecture of the Lower Echelon


TI Including UHF TACSAT as the Range Extension
for the EPLRS backbone
In this paper, the performance of the using the MILSTD 188-184 multi-access scheme over dedicated
channels for TI range extension will be assessed. The
performance of MIL-STD 188-184 over DAMA channels
will not be discussed in this paper. The assessment
presented in this paper is performed in two stages. The
first stage is a brief qualitative assessment. The purpose of
the qualitative assessment is to eliminate certain MIL-STD
188-184 modes, which will not meet the TI throughput and
delay requirements. The second stage is a quantitative
assessment for those MIL-STD 188-184 modes not
eliminated by the qualitative assessment.
Section 2 presents a brief overview of the MIL-STD
188-184 modes.
Section 3 presents the qualitative
discussion on the applicability of the different MIL-STD
188-184 modes in meeting the TI requirements. Section 4
presents the quantitative assessment results. Section 5
presents the summary and conclusions.

0-7803-5538-5/99/$10.00 (c) 1999 IEEE

2. MIL-STD 188-184 Protocol Description


In this section, a description of the MIL-STD 188-184
link layer protocols will be discussed. MIL-STD 188-184
has many modes. MIL-STD 188-184 supports both
unicasting (i.e. Point-to-point) and multicasting (i.e. one to
many). MIL-STD 188-184 also provides link layer
reliability through two mechanisms. The first mechanism
is packet acknowledgements for both unicast and multicast
datagrams. In the case of the multicast datagrams, all
participants are expected to send their packet
acknowledgements. The second reliability mechanism
uses what is termed probing. This mechanism uses
small packets prior to message transmission to ensure
access to the channel. In the probing mode, a small probe
is sent and probe acknowledgement is returned. Upon
receiving the probe ack., this indicates to other users that
the channel captured for message transmission. The
reason that probing improves reliability is because other
users on the network will cease trying to access the
channel upon receiving a probe. The probing mechanism
minimizes interference from other users, which improves
transmission reliability.
But the cost of improving
reliability through probing is the increase in message
delay.
From the above MIL-STD 188-184 summary, there
are therefore, six different modes which are the
combination of unicast/multicast, acknowledged/nonacknowledged, and probed/non-probed. The next few
paragraphs will briefly describe each of those modes.

2.1 Multi-access with no probing or acking


Figure 2a, illustrates the flowchart of the multi-access
point-to-point and multicast protocol with no probing or
acking. As shown in Figure 2a, the protocol first senses
for channel activity. If the channel is busy, then wait until
the channel becomes idle. If the channel is idle then reset
the wait timer (). The purpose of is to allow time for the
acknowledgements of previous messages to be transmitted
prior to other users trying to access the channel. If the
channel becomes busy during , then wait until the channel
becomes idle. If the channel is idle during , then calculate
the backoff time (). The purpose of the is to randomize
the users such that multiple users dont try the access the
channel simultaneously, resulting in a potential message
collision. The is a uniform random number from 0 to 4
secs. If the channel becomes busy during the , then exit
the backoff process and wait until channel becomes idle.
If the channel is idle during the , then the message is
transmitted upon the expiration of . Table 2b illustrates
the various values for . The values of are based on a slot
time of 3035 msec for a data rate of 2.4 Kbps. Notice that
the values of vary based on the type of the previously
transmitted message. Also notice that there are 3 other

values for the . The other values are termed fast,


fastest and instant. In the instant mode, is set to 0
secs. These other values are not part of the MIL-STD 188184. The reason for including additional values of , will
be discussed in section 4.
ADC/IP turn on

Data Rate

Interburst delay

2400
4800

0.5 sec
0.9 sec
3 sec

16000

Reset idle channel


wait timer ()
Calculate backoff time
For first burst, backoff time
is uniform random from 0-4
sec)
For subsequent bursts, the
backoff time is interbrust
delay

End
message
yes

no

Message
contains
multiple bursts

no

Sense Channel

If Channel is
Busy ?

yes
If Channel
becomes busy during
the idle channel wait
time () ?

no

Wait until channel


is idle

yes
no
Transmit
message
burst

If a busy channel
is detected during
the backoff period ?
yes
Exit backoff period
wait for the channel to
become idle and
proceed with step 4

Figure 2a. Multi-access with no Probing or ACKing


Protocol Flowchart
Non-Standard Modes
Burst Type

Fixed WaitTime
() (msec) Normal

Fast

Fastest

Instant

Probe Burst
Probe Ack
Point to point
(not end of message
)

Second slot time-out value


Nine slot time-out value

2
5

2
2

0
0

Nine slot time-out value

PP (end of
message)

Second slot time-out value

Use pervious fixed wait period

Ack
Multicast Probe Burst
Multicast message
(not end of message
Multicast message
(end of

Slot time = destination count +1


Nine slot time-out value

Fourth slot time-out value

Slot time = 3035 msec for 2400 bps

) for the Multi-access


Figure 2b. Wait-time ()
PTP/Multicast without Probing or Acking Mode

2.2 Multi-access with acking and no probing


In this section, we will discuss the protocol for PTP
and multicast with acking and no probing. For the PTP
with acking, accessing the channel is identical to the
previous case discussed in section 2.1. The difference is
when the first segment of the message is transmitted, an
acknowledgment timer () is set by the transmitter. If an
ack is received before the expiration of , then transmit the
next burst and the erroneous packets from the previous
burst if any. If an ack is not received before the expiration
of , and the maximum number of retires is not yet
reached, then retransmit the previous burst. If the max
number of retries have been reached, then the message is
discarded.

0-7803-5538-5/99/$10.00 (c) 1999 IEEE

The protocol of the multicast with acking and no


probing is very similar to the PTP protocol discuss above
but it differs in that all multicast recipients will have to
acknowledge a given burst. Each recipient has a different
depending on the recipients order from the link layer
header of the burst.

3. Discussion
In this section, we will provide a qualitative
assessment of the applicability of the different MIL-STD
188-184 modes based on the TI traffic distribution,
acceptable message delays and message completion rate
requirements.

2.3 Multi-access with acking and probing


In this section, we will discuss the PTP and multicast
protocols with acking and probing. Each of the protocols
is divided into two portions. The first portion deals with
the transmission of the probe packet. The second portion
deals with the transmission of the data message.
First we will discuss the PTP with acking and probing
protocol. The first portion of the protocol deals with the
transmission and acknowledgement of the probe packet.
Accessing the channel for the probe transmission is
identical protocol described in section 2.1. Once the probe
is transmitted, a probe acknowledgement timer () is set.
If an acknowledgement is received prior to the expiration
of the , then the message transmission begins. If
expires prior to receiving an acknowledgement, then the
probe is retransmitted after a backoff period (). Once the
probe process is complete, all other station sharing the
same channel, cease transmission for the duration of the
message transmission. Therefore, the probe process
ensures that when the message is transmitted, there is no
interference from other terminals on the channel. In
addition, the probe packets are used to reserve the channel
in lieu of using the message. The use of the probe packets
for channel reservation improves the channel utilization
because the probe packets are small in size. The message
transmission protocol is identical to the PTP with ack
described in section 2.2.
Now let us discuss multicast with acking and probing
protocol. The protocol is again divided into two portions.
The first portion guarantees the successful transmission of
the multicast probe packet. The second portion guarantees
the successful transmission of the multicast message. Once
the probe is transmitted, a probe acknowledgement timer
() is set for each of the multicast destinations. If all
acknowledgements are received prior to the expiration of
the s, then the message transmission begins. If one or
more of the s expire prior to receiving an
acknowledgement, then the probe is retransmitted after a
backoff period () only to those destination that failed to
acknowledge. This process continues until the maximum
number of transmissions has expired.
Multicast
destinations that dont send probe acknowledgement are
excluded from the message destination list. The message
transmission protocol is identical to the multicast with ack
described in section 2.2.

First let us discuss the TI requirements. The Lower


Echelon TI traffic consists of SA and C2. The SA traffic is
broadcast in nature. In another words SA is transmitted to
all destinations in the TI. On the other hands, the majority
of the C2 traffic is multi-destined. C2 traffic that is only
uni-destined represents a very small portion of the overall
traffic. The desired TI message delay is assumed to be on
the order of a few seconds (i.e. 4-15 secs) for both SA and
C2. The desired TI message completion rate is assumed to
be at least 80 % for SA and 80-85 % for C2 multicast.
Now let us revisit how the UHF TAC-SAT is
envisioned to be architected into the Lower Echelon TI.
As shown in Figure 1, the UHF TAC-SAT is used to
extend the TI lower echelon backbone (i.e. EPLRS).
Because of the limited UHF TAC-SAT resources (i.e.
number of channels and channel capacity), the range
extension is only used for a limited number of units (i.e.
BDE recon) in the first digitized division (FDD). The
range extension architecture should handle the broadcast of
SA over TAC-SAT as well as the transmission of C2
multicast and unicast. For C2 multicast, the architecture
should not limit the group membership. For that reason,
multicast messages transmitted over the TAC-SAT are
treated as a broadcast on the link layer. The multicast
group members would filter their participation on the IP
layer. Therefore, it can be easily seen that for SA and C2
multicast messages, link layer broadcast is sufficient. For
C2 unicast, the link layer is PTP. Therefore, as far as the
MIL-STD 188-184 is concerned, the desired modes of
operation are the PTP for IP unicast traffic, and the
broadcast for IP multicast traffic.
From these
requirements, the need for the MIL-STD 188-184
multicast capability is eliminated. Now the question
becomes whether or not to use probing or acking or both.
Since the message delay requirement is assumed to be
between 4-15 secs, this biases the MIL-STD 188-184
mode selection toward the unacknowledged and unprobed
mode as described in section 2.1. The reason for this
selection is because of the excessive delay encountered for
probing and the acking due to the fixed time, the backoff
time, the probing acknowledgement timeout and message
acknowledgment timeout. All these timers cause delays
that exceed the assumed message delay requirements.
Therefore, we must first examine the performance without
acking and probing. In addition, to reduce the delays even
further, the fixed time is also reduced to zero. Remember

0-7803-5538-5/99/$10.00 (c) 1999 IEEE

90
80

30

25

70

20

60
50

15

40
10

30

Average Delay (sec)

4. Simulation Results

Completion Rate - 10 min SA


Completion Rate - 5 min SA
Completion Rate - 2 min SA
Average Delay - 10 min SA
Average Delay - 5 min SA
Average Delay - 2 min SA

100

Completion Rate (%)

that the main function of the fixed time is to allow time for
message acknowledgements. Since we are considering
unack and unprobed message transmission, the fixed time
can be eliminated. The issue now becomes, can we still
achieve the desired message success rates for SA and C2
without using probing, acking and eliminating the fixed
time. A simulation was constructed to address this issue.
The results are presented in the next section.

20
5

2.8

2.6

3.2

3.7

5.1

4.3

6.4

8.5

16.0

12.8

21.3

32.0

64.0

128.0

C2 Packet Interarrival Time (min)

Figure 5a. SA Completion Rate and Delay Performance


with no Fixed Time
Completion Rate - 10 min SA
Completion Rate - 5 min SA
Completion Rate - 2 min SA
Average Delay - 10 min SA
Average Delay - 5 min SA
Average Delay - 2 min SA

90

80

35

30

25

70

60
20
50
15
40

30

10

Average Delay (sec)

100

20
5

2.6

2.8

3.2

3.7

4.3

6.4

5.1

8.5

16.0

12.8

21.3

32.0

64.0

10

128.0

We will first present the sensitivity analysis results for


varying the fixed time on the delay and completion rate
performance. Figures 5a and 5b illustrate the SA and C2
performance with no fixed time. Figures 6a and 6b
illustrate the SA and C2 performance with a fixed time of
2 slots (i.e. each slot is approx. 3 secs). As shown in
Figure 5a and 5b, the three curves represent a SA rate of
2,5 and 10 mins while in Figure 6a, and 6b, the three
curves represent a SA rate of 5,10 and 15 mins. The x-axis
represents the interarrival time for 40 C2 messages during
the indicated periods. The left y-axis represents the
message completion rate while the right y-axis represents
the delay in secs.

Completion Rate (%)

In this section, the simulation results are presented.


There are two cases examined by the simulation. The first
case is a sensitivity analysis for varying the fixed time.
The second case is a sensitivity analysis for varying the
backoff time.

10

C2 Packet Interarrival Time (min)

Figure 5b. C2 Completion Rate and Delay Performance


with no Fixed Time
Completion Rate - 15 min SA
Completion Rate - 10 min SA
Completion Rate - 5 min SA

100

80

Average Delay - 15 min SA

90

70

Average Delay - 10 min SA


Average Delay - 5 min SA

80

60
70
50
60

50

40

40
30
30
20
20

Average Delay (sec)

Completion Rate (%)

Results indicate that to achieve an 85 % SA and C2


completion rate, without the fixed time, the supportable
SA rate is 5 mins, with C2 interarrival time of 100 secs, a
SA delay of < 5 secs and a C2 delay of 7 secs. On the
other hand, with the fixed time, the supportable SA rate is
10 mins, with C2 interarrival time of 3 min, a SA delay of
8 secs and a C2 delay of 10 secs. This result indicates that
having a fixed time decreases the channel overall
utilization. The reason for that is having a fixed time
increases the queue backlog, resulting in more collision
and thus reducing the overall channel utilization and
increasing the message delays.

10

2.6

2.8

3.2

3.7

4.3

5.1

6.4

8.5

12.8

16.0

21.3

32.0

64.0

128.0

10

C2 Packet Interarrival Time (min)

Figure 6a. SA Completion Rate and Delay Performance


with a Fixed Time = 6 secs

0-7803-5538-5/99/$10.00 (c) 1999 IEEE

Completion Rate - 15 min SA


90

Completion Rate - 10 min SA

100

Completion Rate - 5 min SA


Average Delay - 15 min SA

90

80

160

90

140

70

20

2.6

2.8

3.2

3.7

4.3

5.1

6.4

8.5

12.8

16.0

21.3

32.0

64.0

128.0

We will now present the sensitivity analysis results for


varying the backoff time. Figures 7a and 7b illustrate the
SA and C2 performance with no fixed time and increasing
the backoff time from a max of 4 secs to 16 secs.
Comparing the results in Figures 7a and 7b, and the
Figures 5a and 5b, yield that increasing the upper limit of
the backoff timer from 4 to 16 secs improves the
completion rates at higher C2 loading. But this is on the
expense of delay. Results indicate that to achieve an 85 %
SA and C2 completion rate, without the fixed time, the
supportable SA rate is 5 mins, with C2 interarrival time of
100 secs, a SA delay of < 5 secs and a C2 delay of 7 secs.
On the other hand, with the increase backoff time, the
supportable SA rate is 2 mins, with C2 interarrival time of
20 sec, a SA delay of 20 secs and a C2 delay of 25 secs.
This result indicates that increasing the backoff increases
the throughput. But increasing the throughput yields an
increase in the message delay as illustrated from the results
above.
160

140

120
70
100
Completion Rate - 10 min SA
Completion Rate - 5 min SA

50

80

Completion Rate - 2 min SA


Average Delay - 10 min SA

40

60

Average Delay - 5 min SA


Average Delay - 2 min SA

30

40

Average Delay (sec)

Completion Rate (%)

80

60

Average Delay - 2 min SA


60

30
40
20
20

2.6

3.2

2.8

3.7

4.3

5.1

6.4

8.5

12.8

C2 Packet Interarrival Time (min)

Figure 6b. C2 Completion Rate and Delay Performance


with a Fixed Time = 6 secs

90

80

Average Delay - 5 min SA

40

C2 Packet Interarrival Time (min)

100

50

10

Average Delay - 10 min SA

10

20

10

100

Completion Rate - 2 min SA

16.0

30
30

Completion Rate - 5 min SA


60

21.3

40
40

Completion Rate - 10 min SA

32.0

50
50

120
70

64.0

60
60

80

128.0

70

Completion Rate (%)

Average Delay - 5 min SA

Average Delay (sec)

Completion Rate (%)

Average Delay - 10 min SA


80

Average Delay (sec)

100

Figure 7b. C2 Completion Rate and Delay Performance


with no Fixed Time and Varying the Backoff Timer

5. Summary/Conclusion
In this paper, a performance assessment of multiaccess over dedicated UHF TACSAT channels using the
MIL-STD 188-184 was conducted. First a qualitative
assessment of the applicability of the MIL-STD 188-184
various modes to the TI range extension was conducted.
From the qualitative assessment, it was first concluded that
the MIL-STD 188-184 multicast mode could be
eliminated. SA and C2 multicast messages will utilize a
broadcast link layer instead for easy implementation.
Secondly, it was also concluded that because of the
required message delay constraint, only the unack and
unprobed modes are evaluated. A quantitative assessment
was then conducted to determine the delay and the success
rate sensitivity to the fixed time and the backoff time. The
assessment yielded that no fixed time and the increase of
the backoff parameter improves the data throughput and
reliability with an acceptable increase in delay.
Based on the results above, two recommendations to
the MIL-STD 188-184 can be made, which will improve
the message performance for the TI applications. The first
recommendation is to allow the fixed time to be settable to
different values other than just the default. Recommended
values are shown in Figure 2b.
The second
recommendation is to increase the upper bound of the
backoff parameter and allow the user to tune it depending
on the application.

References

20
20

2.6

0
2.8

3.7

4.3

5.1

6.4

8.5

12.8

16.0

21.3

32.0

64.0

128.0

3.2

10

C2 Packet Interarrival Time (min)

[1] MIL-STD 188-184, Interoperability and Performance


standard for data control waveform, August 1993
[2] Yosry Barsoum, and Gus Amouris, UHF TACSAT
integration into the TI, MITRE Corp., 1998.

Figure 7a. SA Completion Rate and Delay Performance


with no Fixed Time and Varying the Backoff Timer

0-7803-5538-5/99/$10.00 (c) 1999 IEEE

You might also like